DESCRIPTION: The proposed project will use longitudinal data from the Health and Retirement Survey (HRS) to examine the determinants of retirement behavior of survey participants. A structural, dynamic model of retirement that takes into account the effects of stochastic events on the need to reoptimize behavior will be estimated. The opportunity set will reflect earnings opportunities as well as the effects of work. Earnings opportunities will be estimated from self-reported data and from earnings records obtained from the Social Security Administration. Incentives from pension plans affecting retirement will be estimated using employer provided pension plan descriptions. Retirement incentives from social security are based on the benefit formula. Health status will be estimated from indices reported in detail in the survey. Indices of working conditions and pressures to retire as well as measures of time preference and risk aversion will be included as determinants of the value of leisure time. Each model will allow for heterogeneity in preferences. The empirical results will be used to examine a number of issues that have been of concern to retirement researchers, including the inability of some previous models to explain the peak in the retirement hazard at age 65. The investigators will also compare results using the HRS with those from the Survey of Consumer Finances and the National Longitudinal Survey of Mature Women.